INDEX

Explanations

manipulative and deceptive activities in political or social contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

collusion and covert operations

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 transfieras

-0.62

 consumidor

-0.53

MemoryWarning

-0.52

balleur

-0.52

paravant

-0.49

 المعيارى

-0.47

 anspruchs

-0.47

 escritora

-0.46

 règne

-0.46

 Fußballspieler

-0.46

POSITIVE LOGITS

AlterField

0.42

 collusion

0.40

 defaultstate

0.37

ostro

0.36

 COLL

0.35

Coll

0.34

ôtel

0.33

허

0.33

thenReturn

0.33

 Coll

0.33

Activations Density 0.056%

manipulative and deceptive activities in political or social contexts

collusion and covert operations

No Comments

No Known Activations